AI Is Atomising Culture Into Personalised Niches of One

The year is 1983. At 8pm sharp on 28 February, 106 million viewers tune in to watch the M*A*S*H finale. The next day, everyone from truck drivers to corporate attorneys had the same reference points. For years, M*A*S*H functioned as a national ritual, allowing families across the country to laugh at the same jokes and debate the same plot twists. Of course, it wasn’t just M*A*S*H. Everyone could assume that nearly everyone else had seen shows like Seinfeld or Dallas. In France, Bernard Pivot’s Apostrophes, a literary talk-show, regularly reached six million viewers, or ten percent of the population. This anchored social life in common references and fostered a sense of collective identity. This logic could also be applied to cinema (the Oscar ceremony attracted 85 million viewers in 1973), music (Michael Jackson, Madonna, Prince, Springsteen, and Queen appealed across demographics), video games (entire generations grew up with Tetris, Zelda, Pokemon, or Super Mario) or even software (everyone used the few available corporate tools).

Today, that sense of shared cultural experience has eroded and been replaced by a landscape of personalised content that appeals to ever-diminishing niches. In the US, the number of channels grew from three in the 1950s to hundreds today, with dozens of specialised channels (MTV for music enthusiasts, ESPN for sports fans, and so on). In 1998, the Seinfeld finale reached 76 million viewers; seven years later, the CSI season five finale “Grave Danger” drew at least thirty million; by 2019, the top-rated show (The Big Bang Theory) finale managed just eighteen million. Over these decades, the Oscars lost three quarters of their audience. Internet culture and streaming services simply accelerated this fragmentation, with millions of viewing hours now catering to sociological subgroups on Netflix or Disney+, while various segments of Gen Z doomscroll on very different TikTok or Instagram feeds. The same applies to music (184 million tracks were streamed in 2023, 86 percent of which recorded 1,000 plays or fewer), video games (players are clustered across Twitch communities with little cultural overlap), and software (thousands of verticalised SaaS have replaced the old universal desktop suite).

Many essays have analysed and attempted to explain the fragmentation of culture, and these explanations point to things like the primacy of individualism and the rise of identity politics. The main culprit, however, is probably technology. When the costs of producing and distributing content are high, it is only worth producing content that appeals to a mass audience. When technology reduces those costs, creators can serve niches. You’ll only green-light the production of a high-budget movie if you hope to attract millions to the cinema, but you’ll be happy to produce a TikTok video if it can reach a hundred people. You’ll need, say, a million clients to break even with software that costs ten million dollars to develop, but with costs reduced tenfold, you can address a user base ten times narrower. You’ll have no incentive to appeal to the long tail of music amateurs if distributing a soundtrack implies costly physical manufacturing, but once it becomes possible to upload a song to Spotify, you can earn a living by satisfying a micro-audience craving a micro-genre.

In other words, when creation is expensive, it is scarce and audiences converge on the few things that do exist. Common culture emerges. As creation becomes cheaper, audiences fragment. Now what if creation wasn’t cheap, but free? With AI, creation does become free. Now, for content to be worth producing, it only needs to appeal to one person. AI gives us cultural niches of one. Culture is fragmented; it is about to become atomised.